How to check out test code once, then use it to run multiple tests

Question

I want to run a set of regression tests (built in Selenium and driven by NUnit) using Bamboo. As there are quite a few of them, and I want the results of every test to show up on the wallboard separately, I need to configure a plan for each of the tests (right?). But I do not want to have to perform a checkout / package restore / compilation sequence for my test code for every plan.

Is there a way to do a single checkout / package restore / compilation action first, then run several plans separately (either in sequence or parallel) using the test binaries that have been created?

Kalle Sirkesalo · Answer

Hello, Yes you can actually also comment on an answer.

That's the normal way of doing it with any kind of environment. I just assumed you might want to run it as all bamboo like in your question.

Bas Dijkstra · Answer

Hello Kalle,

thank you for your answer. I have found what I think is a workable solution by:

* Defining a plan that checks out the code, compiles it and stores the binaries into an artifact
* Importing this artifact in subsequent plan that run the various test scenarios
* Defining these test plans as dependencies of the checkout / compile, meaning they are triggered automatically after that one finishes.

Kalle Sirkesalo · Answer

Hello,

For one you can use shallow clones to lower the amount of data you get per fetch.

For another you can cache the data on your repository so it doesn't fetch all the data again.

Then you could just define your results in different folders set them as artifacts and run all the tests in one job for example. This way all of the reports would come in one artifact page and you could run all the test in one job. This is very useful if you want to run the same things on your branches. This way you can just use bamboo branches in their full effect.